Skip to main content

AppMod Blueprints Builder Skill

An AI coding assistant skill that helps you build AWS CDK applications using AppMod Catalog Blueprints.

What It Does​

This skill connects your AI assistant to the AppMod Catalog Blueprints MCP server, enabling it to:

  • Discover constructs — browse available CDK constructs and their props
  • Generate code — scaffold constructs with correct imports, security defaults, and type-safe props
  • Compose applications — wire multiple constructs together with automatic dependency ordering
  • Find examples — locate reference implementations for your use case

Installation​

Claude Code​

1. Add the MCP server:

claude mcp add appmod-catalog-blueprints \
  -- uvx --from "git+https://github.com/cdklabs/cdk-appmod-catalog-blueprints.git@main#subdirectory=mcp-appmod-catalog-blueprints" mcp-appmod-catalog-blueprints

2. Install the skill:

mkdir -p ~/.claude/skills
cp -r skills/appmod-blueprints-builder ~/.claude/skills/

3. Restart Claude Code

Kiro​

1. Add the MCP server to ~/.kiro/settings/mcp.json (global) or .kiro/settings/mcp.json (workspace):

{
  "mcpServers": {
    "appmod-catalog-blueprints": {
      "command": "uvx",
      "args": [
        "--from",
        "git+https://github.com/cdklabs/cdk-appmod-catalog-blueprints.git@main#subdirectory=mcp-appmod-catalog-blueprints",
        "mcp-appmod-catalog-blueprints"
      ]
    }
  }
}

2. Install the skill:

mkdir -p ~/.kiro/skills
cp -r skills/appmod-blueprints-builder ~/.kiro/skills/

3. Add a steering doc to your CDK project to activate the skill:

mkdir -p .kiro/steering
curl -o .kiro/steering/appmod-blueprints-skill.md \
  https://raw.githubusercontent.com/cdklabs/cdk-appmod-catalog-blueprints/main/skills/appmod-blueprints-builder/sample-kiro-steering.md

Or copy from a local clone:

cp path/to/cdk-appmod-catalog-blueprints/skills/appmod-blueprints-builder/sample-kiro-steering.md \
  .kiro/steering/appmod-blueprints-skill.md

Why a steering doc? Ensure Kiro auto-triggers the skills from the SKILL.md description. The steering doc with inclusion: always tells Kiro to activate the skill whenever you're working on relevant topics (CDK, chatbots, AI agents, etc.).

Codex​

1. Add the MCP server to your Codex MCP configuration file:

{
  "mcpServers": {
    "appmod-catalog-blueprints": {
      "command": "uvx",
      "args": [
        "--from",
        "git+https://github.com/cdklabs/cdk-appmod-catalog-blueprints.git@main#subdirectory=mcp-appmod-catalog-blueprints",
        "mcp-appmod-catalog-blueprints"
      ]
    }
  }
}

2. Install the skill:

mkdir -p ~/.codex/skills
cp -r skills/appmod-blueprints-builder ~/.codex/skills/

3. Restart Codex

Usage​

Just describe what you want to build. The skill triggers automatically on keywords like:

  • chatbot, AI agent, interactive agent, batch agent
  • document processing, classification, extraction
  • RAG, knowledge base, Bedrock
  • CDK, serverless AI

Example Prompts​

You SayWhat Happens
"I want to build a customer support chatbot"Asks clarifying questions, then scaffolds InteractiveAgent + Frontend
"Create a document processing pipeline"Composes Network + BedrockDocumentProcessing with proper wiring
"What constructs are available?"Lists all construct families from the MCP catalog
"Show me chatbot examples"Finds example applications using InteractiveAgent

Workflow​

  1. You describe your goal — "Build a chatbot with RAG"
  2. Skill asks clarifying questions — "Do you need a frontend? What kind of knowledge base?"
  3. Skill discovers relevant constructs — queries the MCP server
  4. Skill generates code — scaffolds or composes constructs
  5. Skill explains what to customize — highlights required values and next steps

Available Constructs​

CategoryConstructs
AgentsInteractiveAgent, BatchAgent, BedrockKnowledgeBase
Document ProcessingBedrockDocumentProcessing, AgenticDocumentProcessing
FoundationNetwork, EventbridgeBroker, AccessLog
WebappFrontend (CloudFront + S3)
UtilitiesCloudWatchTransactionSearch, DataLoader

Example Output​

When you ask to "build a chatbot with RAG and a frontend", the skill generates:

import { BedrockKnowledgeBase, Frontend, InteractiveAgent } from '@cdklabs/cdk-appmod-catalog-blueprints';

const kb = new BedrockKnowledgeBase(this, 'KnowledgeBase', {
  knowledgeBaseId: '<YOUR_KB_ID>',
  name: 'support-docs',
  description: 'Product documentation for customer questions',
});

const agent = new InteractiveAgent(this, 'ChatAgent', {
  agentName: 'support-bot',
  agentDefinition: {
    bedrockModel: { useCrossRegionInference: true },
    systemPrompt: new Asset(this, 'Prompt', { path: './prompt.txt' }),
    knowledgeBases: [kb],
  },
  enableObservability: true,
});

const frontend = new Frontend(this, 'Frontend', {
  sourceDirectory: './frontend/dist',
});

Troubleshooting​

IssueSolution
Skill doesn't triggerCheck MCP server is running — restart your AI tool
"MCP tools unavailable"Run the MCP server setup command for your tool
Outdated construct propsClear MCP cache: uv cache prune

Learn More​